Gangnam Style AI prompts
Copy-paste prompts for the Gangnam Style AI look: three people singing and dancing on a terrace at night, with the close-ups, whip pan and solo moments of the original clip. They work in Kling, Veo, Sora and most other video models.
Updated September 25, 2026
The base prompt
This prompt covers the whole clip: the night terrace, three people, the camera and each beat in order. Paste it as is, or replace "Three people" with a short description of your own group.
Three people stand side by side, left, center and right, facing the camera. Setting: a terrace at night with potted plants and old lamp posts, a lit glass building behind the right side of the frame. Cinematic 16:9 film look, a song-and-dance number. The camera moves fast between the three, with quick whip pans. No captions, no subtitles, no text, no logos. Beat by beat: first a close-up of the center person singing into the lens while the other two lean in behind (0-4 s). Then a whip pan to the left person, who sings and gestures alone (4-8 s). Then the right person sings and dances alone in a wider shot (8-12 s). Then back to the center person in close-up, singing and waving an open hand at the lens (12-15 s).
For faces that look like real people, use image-to-video: give the model a photo of your three people standing side by side, and let the prompt handle the setting, the camera and the moves.
Variants
Each variant is a complete prompt, ready to copy. The camera and the beats stay the same; the people and the setting change, like in the popular remakes.
Stadium at night
Football kits on a floodlit pitch, like many of the popular remakes. Describe the kits, not a real team.
Three football players in matching plain kits stand side by side, left, center and right, facing the camera. Setting: a football pitch at night under bright floodlights, empty stands blurred behind them. Cinematic 16:9 film look, a song-and-dance number. The camera moves fast between the three, with quick whip pans. No captions, no subtitles, no text, no logos. Beat by beat: first a close-up of the center person singing into the lens while the other two lean in behind (0-4 s). Then a whip pan to the left person, who sings and gestures alone (4-8 s). Then the right person sings and dances alone in a wider shot (8-12 s). Then back to the center person in close-up, singing and waving an open hand at the lens (12-15 s).
Night rooftop
Matching team jackets with city lights behind. Keep the jackets plain, without real logos.
Three people in matching plain team jackets stand side by side, left, center and right, facing the camera. Setting: a rooftop terrace at night, city lights and tall buildings glowing behind them. Cinematic 16:9 film look, a song-and-dance number. The camera moves fast between the three, with quick whip pans. No captions, no subtitles, no text, no logos. Beat by beat: first a close-up of the center person singing into the lens while the other two lean in behind (0-4 s). Then a whip pan to the left person, who sings and gestures alone (4-8 s). Then the right person sings and dances alone in a wider shot (8-12 s). Then back to the center person in close-up, singing and waving an open hand at the lens (12-15 s).
Castle backdrop
Robes and a stone castle for a fantasy-film look. Make the characters your own rather than copying existing ones.
Three people in long dark robes and scarves stand side by side, left, center and right, facing the camera. Setting: a stone castle with tall towers behind them, at dusk. Cinematic 16:9 film look, a song-and-dance number. The camera moves fast between the three, with quick whip pans. No captions, no subtitles, no text, no logos. Beat by beat: first a close-up of the center person singing into the lens while the other two lean in behind (0-4 s). Then a whip pan to the left person, who sings and gestures alone (4-8 s). Then the right person sings and dances alone in a wider shot (8-12 s). Then back to the center person in close-up, singing and waving an open hand at the lens (12-15 s).
Three friends
The everyday version. Swap in your own outfits and setting.
Three friends in casual streetwear, one in a hoodie, one in a denim jacket and one in a T-shirt, stand side by side, left, center and right, facing the camera. Setting: a city street in the evening, shop lights behind them. Cinematic 16:9 film look, a song-and-dance number. The camera moves fast between the three, with quick whip pans. No captions, no subtitles, no text, no logos. Beat by beat: first a close-up of the center person singing into the lens while the other two lean in behind (0-4 s). Then a whip pan to the left person, who sings and gestures alone (4-8 s). Then the right person sings and dances alone in a wider shot (8-12 s). Then back to the center person in close-up, singing and waving an open hand at the lens (12-15 s).
Family
Siblings or parents in everyday clothes. The center spot gets the close-up, so choose who goes there.
Three adult family members in everyday clothes stand side by side, left, center and right, facing the camera. Setting: a living room with warm lamp light. Cinematic 16:9 film look, a song-and-dance number. The camera moves fast between the three, with quick whip pans. No captions, no subtitles, no text, no logos. Beat by beat: first a close-up of the center person singing into the lens while the other two lean in behind (0-4 s). Then a whip pan to the left person, who sings and gestures alone (4-8 s). Then the right person sings and dances alone in a wider shot (8-12 s). Then back to the center person in close-up, singing and waving an open hand at the lens (12-15 s).
Coworkers
Office clothes and a song-and-dance number between the desks make the joke land.
Three coworkers in office clothes, one in a button-up shirt, one in a blazer and one in a sweater, stand side by side, left, center and right, facing the camera. Setting: an open-plan office with desks and monitors behind them. Cinematic 16:9 film look, a song-and-dance number. The camera moves fast between the three, with quick whip pans. No captions, no subtitles, no text, no logos. Beat by beat: first a close-up of the center person singing into the lens while the other two lean in behind (0-4 s). Then a whip pan to the left person, who sings and gestures alone (4-8 s). Then the right person sings and dances alone in a wider shot (8-12 s). Then back to the center person in close-up, singing and waving an open hand at the lens (12-15 s).
Negative prompt & settings
If your tool has a negative prompt field, paste this in to keep extra people, text and a locked-off camera out of the shot.
more than three people, fourth person, crowd in the foreground, static tripod shot, locked-off camera, slow motion, text, captions, subtitles, logo, watermark, blurry faces, merged faces, distorted hands, extra fingers
- Aspect ratio
- 16:9, the same as the generator's output
- Length
- 15 seconds for the four beats
- Quality
- Standard (std) for drafts, Pro for the final render
- Start image
- A photo of your three people side by side, if the model supports image-to-video
- Camera
- Moving camera with a fast pan, if the tool has camera control. Avoid static or fixed.
Why prompts alone don't nail the beats
A prompt describes a clip; it can't hand over choreography. The model invents its own timing on every run, so you may get three people singing on a terrace, but rarely the four beats in the right order and on the music.
A tool built for one format gets around this by keeping the beats fixed and only changing the people. That is what the one-click generator does: three photos in, the same beats out, with the original audio. The how-to guide compares all three methods side by side.
Prompt FAQ
Which AI video model works best with these prompts?
Any current model that takes a text prompt will render the scene, including Kling, Veo and Sora. Starting from an image of your three people (image-to-video) keeps their faces far closer than text alone.
Can I use these prompts for free?
Yes. Copy them and adapt them however you like. Whether generating the video is free depends on the tool and plan you use.
Why does my result never match the trend clip?
Text prompts don't carry choreography, so the model makes up its own moves and camera path on every run. To get the same beats every time you need a tool that keeps them fixed, like the one-click generator.
Ready to put your three in the clip?
Upload three photos, press Generate video, and get a 15-second MP4 with the original audio in about 10 minutes.